90 Chapter 4 A logistic regression analysis was performed using baseline demographics (age, gender, nationality, education) and clinical (BMI, BED diagnosis, lifetime vomiting, lifetime laxatives, lifetime diuretics, lifetime excessive exercise, OBE episodes, EDE-Q Global, DEBQ Emotional Eating, SCL-90 total, BDI-II total, EDI-3 Low Self-Esteem, EDI-3 Emotional Dysregulation) characteristics to predict treatment assignment. Duration of illness and living situation were not included in the propensity analysis due to the amount of missing data (23/175 and 22/175 respectively). The predicted probability of receiving CBT, referred to as a propensity score, was derived for each study participant from this model. The propensity score was used as a covariate in all outcome analyses to control for pre-treatment differences among the non-randomized groups. Generalized linear models were used to compare treatment groups on primary and secondary measures of outcome at baseline, end of treatment, and follow-up. Primary measures of outcome used to evaluate effectiveness included OBE episodes and EDE-Q Global scores. Secondary measures of outcome included DEBQ Emotional Eating, EDI-3 Emotional Dysregulation, SCL-90 total score, BDI-II total score, and EDI-3 Low Self-Esteem. A negative binomial model with log link (appropriate for count data) was used for OBE episodes. A normal distribution with log link was used for symmetrically distributed measures (EDE-Q Global, DEBQ Emotional Eating, EDI-3 Low Self-Esteem), while a gamma distribution with log link was used for the remaining outcome measures (BDI-II, EDI-3 Emotional Dysregulation, SCL-90 total). Models included a random intercept and fixed effects for treatment, study visit, treatment-by-visit interaction, and propensity score. Treatment outcome at each post-baseline visit was compared by calculating the 95% confidence interval of the difference between treatments and corresponding effect size. Confidence intervals that do not contain zero were considered evidence of a significant difference in outcome between treatments. Given that treatment for both CBT and DBT was delivered in group settings, preliminary models were run nesting participants within therapists within therapeutic groups. As no significant variation attributable to therapist or therapeutic group was found, subsequent analyses were conducted without nesting. Effect sizes between treatments were calculated using both Cohen’s (Cohen, 1988) d and the success rate difference (SRD; Kraemer & Kupfer, 2006). Cohen’s d values were calculated from covariate-adjusted estimated marginal means; Cohen uses values of 0.2, 0.5, and 0.8 to characterize “small”, “medium”, and “large” differences between groups, respectively. SRD values, which can range from − 1 to + 1, represent the probability that a randomly selected case from one treatment will have an outcome preferable to a randomly selected case from another treatment.
RkJQdWJsaXNoZXIy MjY0ODMw